Identification and replication of susceptibility genes for Parkinson disease at the population level have been hampered by small studies with potential biases. Alpha-synuclein (SNCA) has been one of the most promising susceptibility genes, but large-scale studies have been lacking. To determine whether allele-length variability in the dinucleotide repeat sequence (REP1) of the SNCA gene promoter is associated with Parkinson disease susceptibility, whether SNCA promoter haplotypes are associated with Parkinson disease, and whether REP1 variability modifies age at onset. We performed a collaborative analysis of individual-level data on SNCA REP1 and flanking markers in patients with Parkinson disease and controls. Study site recruitment, data collection, and analyses were performed between April 5, 2004, and December 31, 2005. Eighteen participating sites of a global genetics consortium provided clinical data. Genotyping was performed for SNCA REP1, -770, and -116 markers at individual sites; however, each site also provided 20 DNA samples for regenotyping centrally. Measures included estimations of Hardy-Weinberg equilibrium in controls; a test of heterogeneity; analyses for association of single variants or haplotypes; and survival analyses for age at onset. Of the 18 sites, 11 met stringent criteria for concordance with Hardy-Weinberg equilibrium and low genotyping error rate. These 11 sites provided complete data for 2692 cases and 2652 controls. There was no heterogeneity across studies (P>.60). The SNCA REP1 alleles differed in frequency for cases and controls (P<.001). Genotypes defined by the 263 base-pair allele were associated with Parkinson disease (odds ratio, 1.43; 95% confidence interval, 1.22-1.69; P<.001 for trend). Multilocus haplotypes differed in frequency for cases and controls (global score statistic, P<.001). Two-loci haplotypes were associated with Parkinson disease only when they included REP1 as one of the loci. However, genotypes defined by REP1 alleles did not modify age at onset (P = .55). This large-scale collaborative analysis demonstrates that SNCA REP1 allele-length variability is associated with an increased risk of Parkinson disease.